Why have the national media and pollsters avoiding New Hampshire?
In 2008, after polling behind Hillary Clinton in New Hampshire for about 360 of the previous 365 days, Barack Obama finished behind Clinton in the New Hampshire primary. The result was shocking to people who viewed the race only in the context of the week between the Iowa Caucuses and the New Hampshire primary, with some even suggesting fraud. There was no fraud, it was just what had been a solid Hillary lead for a year shrinking to a few points due to Obama’s momentum out of Iowa, with the votes distributed much like they were between Kerry & Dean.
The polling right after Iowa showed Obama surging. When he didn’t, pollsters were confused and even a bit embarrassed. The more scrupulous went back to their data and tried to figure out what went wrong. There was no single accepted answer, but rather a grab bag of partial explanations: Obama’s performance in the debate between Iowa and New Hampshire, Clinton’s tears in New Hampshire, independents who leaned toward Obama deciding to hedge their bets and shifting to the Republican primary to vote for McCain, and one I’m partial to, that the targeted field effort by EMILY’s List, AFSCME and the American Federation of Teachers helped Clinton over-perform among women voters.
While there were plenty of reasonable explanations for why methodologically sound polls could not predict the final results, New Hampshire gave polling a public black eye. It shouldn’t have; the generally low quality of most public polls didn’t begin in January 2008. But one has to wonder if national media has decided against putting their names–and thus, their reputations–on polls that they could worry would be wrong and again discredit their results. Based on the run-up to today’s New Hampshire primary, it looks that way.
In 2008, according to the Real Clear Politics poll tracker, there were 25 polls of the New Hampshire Republican primary that were in the field on January 1 or later. [For this discussion I’m not including Rasmussen, whose reputation does not warrant it being considered a professional and reliable pollster.] These 25 polls were conducted by 11 different pollsters, 8 of whom–American Research Group, Zogby, Marist, Fox, Strategic Vision, Gallup, Research 2000 and Mason-Dixon–were firms that did polling around the country. The remaining three–Suffolk, University of New Hampshire and Franklin Pierce–poll primarily in New England.
Now, if you closely follow polling, you’re probably objecting to this list of “reputable” pollsters. You should, as several have dubious reputations. ARG is a mystery firm. They never release data, crosstabs, information on who pays for their polls or anything else. Zogby has a poor reputation among pollsters, both for quality of work and for not revealing potential conflicts of interest. Strategic Vision was pretty much run out of business after being exposed by Nate Silver. And Research2000, which was paid to poll for the Concord Monitor, was also exposed as a fraud.
So, we see one reason why “the polls” were wrong in 2008: a bunch of them were by crappy pollsters, most likely because media outlets liked having poll numbers but didn’t like paying the cost of a real poll by a serious pollster. A few of these firms–ARG, JZ Analytics (a spinoff of Zogby run by a Jonathan Zogby, presumably the son of John Zogby)–are back this year providing numbers (unreliable though they may be). Marist is back, this year for NBC. But Mason-Dixon hasn’t polled NH since October, and Fox and Gallup haven’t polled it at all.
In addition to the national pollsters, the national media haven’t sponsored as many late polls of NH as in 2008. Reuters, C-Span, CNN, Fox & USA Today all sponsored polls in 2008 but not in 2012. Among national media only NBC and (the also dubious) Washington Times have sponsored 2012 polls (the Washington Times using the Zogby spinoff).
In addition to having fewer national media sponsors, in 2012 there are simply fewer polls; after 25 January polls in 2008, this time there have only been 13. Of those thirteen, over half have been by one pollster, Suffolk. And there are also fewer national pollsters. Other than JZ Analytics, ARG, and Marist, Public Policy Polling–which didn’t poll New Hampshire in 2008–is the only national pollster to release a January poll.
Despite having fewer polls from pollsters with sound national reputations, it’s possible the polling we’ve seen from New Hampshire will be consistent with tonight’s results. But if not, it may be because national media companies and national pollsters–for cost cutting reasons, or maybe to avoid embarrassment–decided to not poll the New Hampshire primary.